def test_reaper():
""" REAPER (DAEMON): Test the reaper daemon."""
nb_files = 30
file_size = 2147483648L # 2G
for i in xrange(nb_files):
replica_core.add_replica(rse='MOCK',
scope='data13_hip',
name='lfn' + generate_uuid(),
bytes=file_size,
account='root',
adler32=None,
md5=None)
rse_core.set_rse_usage(rse='MOCK',
source='srm',
used=nb_files * file_size,
free=800L)
rse_core.set_rse_limits(rse='MOCK',
name='MinFreeSpace',
value=10737418240L)
rse_core.set_rse_limits(rse='MOCK', name='MaxBeingDeletedFiles', value=10)
rses = [
rse_core.get_rse('MOCK'),
]
reaper(once=True, rses=rses)
reaper(once=True, rses=rses)
def test_reaper_multi_vo(vo, second_vo, scope_factory, caches_mock, file_config_mock):
""" REAPER (DAEMON): Test the reaper daemon with multiple vo."""
[cache_region] = caches_mock
new_vo = second_vo
_, [scope_tst, scope_new] = scope_factory(vos=[vo, new_vo])
nb_files = 250
file_size = 200 # 2G
rse1_name, rse1_id, dids1 = __add_test_rse_and_replicas(vo=vo, scope=scope_tst, rse_name=rse_name_generator(),
names=['lfn' + generate_uuid() for _ in range(nb_files)], file_size=file_size)
rse2_name, rse2_id, dids2 = __add_test_rse_and_replicas(vo=new_vo, scope=scope_new, rse_name=rse_name_generator(),
names=['lfn' + generate_uuid() for _ in range(nb_files)], file_size=file_size)
rse_core.set_rse_limits(rse_id=rse1_id, name='MinFreeSpace', value=50 * file_size)
rse_core.set_rse_limits(rse_id=rse2_id, name='MinFreeSpace', value=50 * file_size)
# Check we reap all VOs by default
cache_region.invalidate()
rse_core.set_rse_usage(rse_id=rse1_id, source='storage', used=nb_files * file_size, free=1)
rse_core.set_rse_usage(rse_id=rse2_id, source='storage', used=nb_files * file_size, free=1)
both_rses = '%s|%s' % (rse1_name, rse2_name)
reaper(once=True, rses=[], include_rses=both_rses, exclude_rses=None)
reaper(once=True, rses=[], include_rses=both_rses, exclude_rses=None)
assert len(list(replica_core.list_replicas(dids=dids1, rse_expression=both_rses))) == 200
assert len(list(replica_core.list_replicas(dids=dids2, rse_expression=both_rses))) == 200
def test_reaper():
""" REAPER2 (DAEMON): Test the reaper2 daemon."""
nb_files = 30
file_size = 2147483648 # 2G
rse_id = rse_core.get_rse_id(rse='MOCK')
for i in range(nb_files):
replica_core.add_replica(rse_id=rse_id,
scope=InternalScope('data13_hip'),
name='lfn' + generate_uuid(),
bytes=file_size,
account=InternalAccount('root'),
adler32=None,
md5=None)
rse_core.set_rse_usage(rse_id=rse_id,
source='srm',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id,
name='MaxBeingDeletedFiles',
value=10)
reaper(once=True, rses=[], include_rses='MOCK', exclude_rses=[])
reaper(once=True, rses=[], include_rses='MOCK', exclude_rses=[])
def test_reaper(vo, caches_mock, file_config_mock):
""" REAPER (DAEMON): Test the reaper daemon."""
[cache_region] = caches_mock
scope = InternalScope('data13_hip', vo=vo)
nb_files = 250
file_size = 200 # 2G
rse_name, rse_id, dids = __add_test_rse_and_replicas(vo=vo, scope=scope, rse_name=rse_name_generator(),
names=['lfn' + generate_uuid() for _ in range(nb_files)], file_size=file_size)
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=50 * file_size)
assert len(list(replica_core.list_replicas(dids=dids, rse_expression=rse_name))) == nb_files
# Check first if the reaper does not delete anything if no space is needed
cache_region.invalidate()
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=323000000000)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
assert len(list(replica_core.list_replicas(dids=dids, rse_expression=rse_name))) == nb_files
# Now put it over threshold and delete
cache_region.invalidate()
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=1)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
assert len(list(replica_core.list_replicas(dids, rse_expression=rse_name))) == 200
def test_reaper():
""" REAPER (DAEMON): Test the reaper daemon."""
nb_files = 30
file_size = 2147483648 # 2G
rse_id = rse_core.get_rse_id(rse='MOCK')
for i in range(nb_files):
replica_core.add_replica(rse_id=rse_id,
scope='data13_hip',
name='lfn' + generate_uuid(),
bytes=file_size,
account='root',
adler32=None,
md5=None)
rse_core.set_rse_usage(rse_id=rse_id,
source='srm',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id,
name='MaxBeingDeletedFiles',
value=10)
argv = ['--run-once', '--rses', 'MOCK']
main(argv)
# Test the rucio-reaper console script
cmd = 'rucio-reaper ' + ' '.join(argv)
exitcode, out, err = execute(cmd)
print(cmd, out, err)
nose.tools.assert_equal(exitcode, 0)
def test_set_rse_limits(self):
""" RSE (CLIENTS): Test the update of RSE limits."""
nb_files = 30
file_size = 2147483648L # 2G
for file in xrange(nb_files):
replica_core.add_replica(rse='MOCK', scope='data13_hip', name='lfn' + generate_uuid(), bytes=file_size, account='root', adler32=None, md5=None)
rse_core.set_rse_usage(rse='MOCK', source='srm', used=nb_files*file_size, free=800L)
rse_core.set_rse_limits(rse='MOCK', name='MinFreeSpace', value=10737418240L)
rse_core.set_rse_limits(rse='MOCK', name='MaxBeingDeletedFiles', value=10)
rses = [rse_core.get_rse('MOCK'), ]
reaper(once=True, rses=rses)
reaper(once=True, rses=rses)
def test_archive_of_deleted_dids(vo, did_factory, root_account, core_config_mock, caches_mock, file_config_mock):
""" REAPER (DAEMON): Test that the options to keep the did and content history work."""
[reaper_cache_region, _config_cache_region, _replica_cache_region] = caches_mock
scope = InternalScope('data13_hip', vo=vo)
account = root_account
nb_files = 10
file_size = 200 # 2G
rse_name, rse_id, dids = __add_test_rse_and_replicas(vo=vo, scope=scope, rse_name=rse_name_generator(),
names=['lfn' + generate_uuid() for _ in range(nb_files)], file_size=file_size, epoch_tombstone=True)
dataset = did_factory.make_dataset()
did_core.attach_dids(dids=dids, account=account, **dataset)
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=50 * file_size)
assert len(list(replica_core.list_replicas(dids=dids, rse_expression=rse_name))) == nb_files
# Check first if the reaper does not delete anything if no space is needed
reaper_cache_region.invalidate()
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=323000000000)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None, greedy=True)
assert len(list(replica_core.list_replicas(dids=dids, rse_expression=rse_name))) == 0
file_clause = []
for did in dids:
file_clause.append(and_(models.DeletedDataIdentifier.scope == did['scope'], models.DeletedDataIdentifier.name == did['name']))
session = get_session()
query = session.query(models.DeletedDataIdentifier.scope,
models.DeletedDataIdentifier.name,
models.DeletedDataIdentifier.did_type).\
filter(or_(*file_clause))
deleted_dids = list()
for did in query.all():
print(did)
deleted_dids.append(did)
assert len(deleted_dids) == len(dids)
query = session.query(models.DataIdentifierAssociationHistory.child_scope,
models.DataIdentifierAssociationHistory.child_name,
models.DataIdentifierAssociationHistory.child_type).\
filter(and_(models.DataIdentifierAssociationHistory.scope == dataset['scope'], models.DataIdentifierAssociationHistory.name == dataset['name']))
deleted_dids = list()
for did in query.all():
print(did)
deleted_dids.append(did)
assert len(deleted_dids) == len(dids)
def test_reaper():
""" REAPER2 (DAEMON): Test the reaper2 daemon."""
if config_get_bool('common', 'multi_vo', raise_exception=False, default=False):
vo = {'vo': config_get('client', 'vo', raise_exception=False, default='tst')}
else:
vo = {}
rse_name = rse_name_generator()
rse_id = rse_core.add_rse(rse_name, **vo)
mock_protocol = {'scheme': 'MOCK',
'hostname': 'localhost',
'port': 123,
'prefix': '/test/reaper',
'impl': 'rucio.rse.protocols.mock.Default',
'domains': {
'lan': {'read': 1,
'write': 1,
'delete': 1},
'wan': {'read': 1,
'write': 1,
'delete': 1}}}
rse_core.add_protocol(rse_id=rse_id, parameter=mock_protocol)
nb_files = 30
file_size = 2147483648 # 2G
file_names = []
for i in range(nb_files):
file_name = 'lfn' + generate_uuid()
file_names.append(file_name)
replica_core.add_replica(rse_id=rse_id, scope=InternalScope('data13_hip', **vo),
name=file_name, bytes=file_size,
tombstone=datetime.utcnow() - timedelta(days=1),
account=InternalAccount('root', **vo), adler32=None, md5=None)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=800)
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id, name='MaxBeingDeletedFiles', value=10)
if vo:
reaper(once=True, rses=[], include_rses='vo=%s&(%s)' % (vo['vo'], rse_name), exclude_rses=[])
reaper(once=True, rses=[], include_rses='vo=%s&(%s)' % (vo['vo'], rse_name), exclude_rses=[])
else:
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=[])
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=[])
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **vo), 'name': n} for n in file_names], rse_expression=rse_name))) == nb_files - 5
def set_rse_usage(rse, source, used, free, issuer, vo='def'):
"""
Set RSE usage information.
:param rse: The RSE name.
:param source: the information source, e.g. srm.
:param used: the used space in bytes.
:param free: the free space in bytes.
:param issuer: The issuer account.
:param vo: The VO to act on.
:returns: List of RSE usage data.
"""
rse_id = rse_module.get_rse_id(rse=rse, vo=vo)
kwargs = {'rse': rse, 'rse_id': rse_id}
if not permission.has_permission(
issuer=issuer, vo=vo, action='set_rse_usage', kwargs=kwargs):
raise exception.AccessDenied(
'Account %s can not update RSE usage information for RSE %s' %
(issuer, rse))
return rse_module.set_rse_usage(rse_id=rse_id,
source=source,
used=used,
free=free)
def test_reaper():
""" REAPER (DAEMON): Test the reaper daemon."""
if config_get_bool('common',
'multi_vo',
raise_exception=False,
default=False):
vo = {'vo': 'tst'}
else:
vo = {}
nb_files = 30
file_size = 2147483648 # 2G
rse_id = rse_core.get_rse_id(rse='MOCK', **vo)
for i in range(nb_files):
replica_core.add_replica(rse_id=rse_id,
scope=InternalScope('data13_hip', **vo),
name='lfn' + generate_uuid(),
bytes=file_size,
account=InternalAccount('root', **vo),
adler32=None,
md5=None)
rse_core.set_rse_usage(rse_id=rse_id,
source='srm',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id,
name='MaxBeingDeletedFiles',
value=10)
argv = ['--run-once', '--rses', 'MOCK']
main(argv)
# Test the rucio-reaper console script
cmd = 'rucio-reaper ' + ' '.join(argv)
exitcode, out, err = execute(cmd)
print(cmd, out, err)
nose.tools.assert_equal(exitcode, 0)
def set_rse_usage(rse, source, used, free, issuer):
"""
Set RSE usage information.
:param rse: The RSE name.
:param source: the information source, e.g. srm.
:param used: the used space in bytes.
:param free: the free space in bytes.
:param issuer: The issuer account.
:returns: True if successful, otherwise false.
"""
kwargs = {'rse': rse}
if not permission.has_permission(issuer=issuer, action='set_rse_usage', kwargs=kwargs):
raise exception.AccessDenied('Account %s can not update RSE usage information for RSE %s' % (issuer, rse))
return rse_module.set_rse_usage(rse=rse, source=source, used=used, free=free)
def set_rse_usage(rse,
source,
used,
free,
issuer,
files=None,
vo='def',
session=None):
"""
Set RSE usage information.
:param rse: The RSE name.
:param source: the information source, e.g. srm.
:param used: the used space in bytes.
:param free: the free space in bytes.
:param issuer: The issuer account.
:param files: the number of files
:param vo: The VO to act on.
:param session: The database session in use.
:returns: True if successful, otherwise false.
"""
rse_id = rse_module.get_rse_id(rse=rse, vo=vo, session=session)
kwargs = {'rse': rse, 'rse_id': rse_id}
if not permission.has_permission(issuer=issuer,
vo=vo,
action='set_rse_usage',
kwargs=kwargs,
session=session):
raise exception.AccessDenied(
'Account %s can not update RSE usage information for RSE %s' %
(issuer, rse))
return rse_module.set_rse_usage(rse_id=rse_id,
source=source,
used=used,
free=free,
files=files,
session=session)
def set_rse_usage(rse, source, used, free, issuer):
"""
Set RSE usage information.
:param rse: The RSE name.
:param source: the information source, e.g. srm.
:param used: the used space in bytes.
:param free: the free space in bytes.
:param issuer: The issuer account.
:returns: True if successful, otherwise false.
"""
kwargs = {'rse': rse}
if not permission.has_permission(
issuer=issuer, action='set_rse_usage', kwargs=kwargs):
raise exception.AccessDenied(
'Account %s can not update RSE usage information for RSE %s' %
(issuer, rse))
return rse_module.set_rse_usage(rse=rse,
source=source,
used=used,
free=free)
def test_source_avoid_deletion(vo, caches_mock, core_config_mock, rse_factory,
did_factory, root_account, file_factory):
""" Test that sources on a file block it from deletion """
_, reaper_region = caches_mock
src_rse1, src_rse1_id = rse_factory.make_mock_rse()
src_rse2, src_rse2_id = rse_factory.make_mock_rse()
dst_rse, dst_rse_id = rse_factory.make_mock_rse()
all_rses = [src_rse1_id, src_rse2_id, dst_rse_id]
any_source = f'{src_rse1}|{src_rse2}'
for rse_id in [src_rse1_id, src_rse2_id]:
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=1)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=1, free=0)
distance_core.add_distance(src_rse1_id, dst_rse_id, ranking=20)
distance_core.add_distance(src_rse2_id, dst_rse_id, ranking=10)
# Upload a test file to both rses without registering
did = did_factory.random_did()
# Register replica on one source RSE
replica_core.add_replica(rse_id=src_rse1_id,
account=root_account,
bytes_=1,
tombstone=datetime(year=1970, month=1, day=1),
**did)
rule_core.add_rule(dids=[did],
account=root_account,
copies=1,
rse_expression=dst_rse,
grouping='ALL',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)
# Reaper will not delete a file which only has one replica if there is any pending transfer for it
reaper_region.invalidate()
reaper(once=True, rses=[], include_rses=any_source, exclude_rses=None)
replica = next(
iter(replica_core.list_replicas(dids=[did],
rse_expression=any_source)))
assert len(replica['pfns']) == 1
# Register replica on second source rse
replica_core.add_replica(rse_id=src_rse2_id,
account=root_account,
bytes_=1,
tombstone=datetime(year=1970, month=1, day=1),
**did)
replica = next(
iter(replica_core.list_replicas(dids=[did],
rse_expression=any_source)))
assert len(replica['pfns']) == 2
# Submit the transfer. This will create the sources.
submitter(once=True,
rses=[{
'id': rse_id
} for rse_id in all_rses],
partition_wait_time=None,
transfertool='mock',
transfertype='single',
filter_transfertool=None)
# None of the replicas will be removed. They are protected by an entry in the sources table
reaper_region.invalidate()
reaper(once=True, rses=[], include_rses=any_source, exclude_rses=None)
replica = next(
iter(replica_core.list_replicas(dids=[did],
rse_expression=any_source)))
assert len(replica['pfns']) == 2
@transactional_session
def __delete_sources(rse_id, scope, name, session=None):
session.execute(
delete(Source).where(Source.rse_id == rse_id,
Source.scope == scope, Source.name == name))
# Deletion succeeds for one replica (second still protected by existing request)
__delete_sources(src_rse1_id, **did)
__delete_sources(src_rse2_id, **did)
reaper_region.invalidate()
reaper(once=True, rses=[], include_rses=any_source, exclude_rses=None)
replica = next(
iter(replica_core.list_replicas(dids=[did],
rse_expression=any_source)))
assert len(replica['pfns']) == 1
def test_reaper():
""" REAPER2 (DAEMON): Test the reaper2 daemon."""
if config_get_bool('common',
'multi_vo',
raise_exception=False,
default=False):
vo = {
'vo': config_get('client',
'vo',
raise_exception=False,
default='tst')
}
new_vo = {'vo': 'new'}
if not vo_core.vo_exists(**new_vo):
vo_core.add_vo(description='Test',
email='*****@*****.**',
**new_vo)
if not scope_core.check_scope(InternalScope('data13_hip', **new_vo)):
scope_core.add_scope(InternalScope('data13_hip', **new_vo),
InternalAccount('root', **new_vo))
nb_rses = 2
else:
vo = {}
new_vo = {}
nb_rses = 1
mock_protocol = {
'scheme': 'MOCK',
'hostname': 'localhost',
'port': 123,
'prefix': '/test/reaper',
'impl': 'rucio.rse.protocols.mock.Default',
'domains': {
'lan': {
'read': 1,
'write': 1,
'delete': 1
},
'wan': {
'read': 1,
'write': 1,
'delete': 1
}
}
}
nb_files = 30
file_size = 2147483648 # 2G
rse_names = []
all_file_names = []
for j in range(nb_rses):
rse_name = rse_name_generator()
rse_names.append(rse_name)
rse_id = rse_core.add_rse(rse_name, **vo)
rse_core.add_protocol(rse_id=rse_id, parameter=mock_protocol)
if new_vo:
rse_id_new = rse_core.add_rse(rse_name, **new_vo)
rse_core.add_protocol(rse_id=rse_id_new, parameter=mock_protocol)
file_names = []
for i in range(nb_files):
file_name = 'lfn' + generate_uuid()
file_names.append(file_name)
replica_core.add_replica(rse_id=rse_id,
scope=InternalScope('data13_hip', **vo),
name=file_name,
bytes=file_size,
tombstone=datetime.utcnow() -
timedelta(days=1),
account=InternalAccount('root', **vo),
adler32=None,
md5=None)
if new_vo:
replica_core.add_replica(
rse_id=rse_id_new,
scope=InternalScope('data13_hip', **new_vo),
name=file_name,
bytes=file_size,
tombstone=datetime.utcnow() - timedelta(days=1),
account=InternalAccount('root', **new_vo),
adler32=None,
md5=None)
all_file_names.append(file_names)
rse_core.set_rse_usage(rse_id=rse_id,
source='storage',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id,
name='MaxBeingDeletedFiles',
value=10)
if new_vo:
rse_core.set_rse_usage(rse_id=rse_id_new,
source='storage',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id_new,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id_new,
name='MaxBeingDeletedFiles',
value=10)
if not vo:
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
assert len(
list(
replica_core.list_replicas(
dids=[{
'scope': InternalScope('data13_hip', **vo),
'name': n
} for n in all_file_names[0]],
rse_expression=rse_name))) == nb_files - 5
else:
# Check we reap all VOs by default
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
assert len(
list(
replica_core.list_replicas(
dids=[{
'scope': InternalScope('data13_hip', **vo),
'name': n
} for n in all_file_names[0]],
rse_expression=rse_names[0]))) == nb_files - 5
assert len(
list(
replica_core.list_replicas(
dids=[{
'scope': InternalScope('data13_hip', **new_vo),
'name': n
} for n in all_file_names[0]],
rse_expression=rse_names[0]))) == nb_files - 5
# Check we don't affect a second VO that isn't specified
reaper(once=True,
rses=[],
include_rses=rse_names[1],
exclude_rses=[],
vos=['new'])
reaper(once=True,
rses=[],
include_rses=rse_names[1],
exclude_rses=[],
vos=['new'])
assert len(
list(
replica_core.list_replicas(
dids=[{
'scope': InternalScope('data13_hip', **vo),
'name': n
} for n in all_file_names[1]],
rse_expression=rse_names[1]))), nb_files
assert len(
list(
replica_core.list_replicas(
dids=[{
'scope': InternalScope('data13_hip', **new_vo),
'name': n
} for n in all_file_names[1]],
rse_expression=rse_names[1]))), nb_files - 5
def test_multihop_intermediate_replica_lifecycle(vo, did_factory, root_account,
core_config_mock,
caches_mock):
"""
Ensure that intermediate replicas created by the submitter are protected from deletion even if their tombstone is
set to epoch.
After successful transfers, intermediate replicas with default (epoch) tombstone must be removed. The others must
be left intact.
"""
src_rse1_name = 'XRD1'
src_rse1_id = rse_core.get_rse_id(rse=src_rse1_name, vo=vo)
src_rse2_name = 'XRD2'
src_rse2_id = rse_core.get_rse_id(rse=src_rse2_name, vo=vo)
jump_rse_name = 'XRD3'
jump_rse_id = rse_core.get_rse_id(rse=jump_rse_name, vo=vo)
dst_rse_name = 'XRD4'
dst_rse_id = rse_core.get_rse_id(rse=dst_rse_name, vo=vo)
all_rses = [src_rse1_id, src_rse2_id, jump_rse_id, dst_rse_id]
did = did_factory.upload_test_file(src_rse1_name)
# Copy replica to a second source. To avoid the special case of having a unique last replica, which could be handled in a special (more careful) way
rule_core.add_rule(dids=[did],
account=root_account,
copies=1,
rse_expression=src_rse2_name,
grouping='ALL',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)
submitter(once=True,
rses=[{
'id': rse_id
} for rse_id in all_rses],
partition_wait_time=None,
transfertype='single',
filter_transfertool=None)
replica = __wait_for_replica_transfer(dst_rse_id=src_rse2_id, **did)
assert replica['state'] == ReplicaState.AVAILABLE
rse_core.set_rse_limits(rse_id=jump_rse_id, name='MinFreeSpace', value=1)
rse_core.set_rse_usage(rse_id=jump_rse_id,
source='storage',
used=1,
free=0)
try:
rule_core.add_rule(dids=[did],
account=root_account,
copies=1,
rse_expression=dst_rse_name,
grouping='ALL',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)
# Submit transfers to FTS
# Ensure a replica was created on the intermediary host with epoch tombstone
submitter(once=True,
rses=[{
'id': rse_id
} for rse_id in all_rses],
partition_wait_time=None,
transfertype='single',
filter_transfertool=None)
request = request_core.get_request_by_did(rse_id=jump_rse_id, **did)
assert request['state'] == RequestState.SUBMITTED
replica = replica_core.get_replica(rse_id=jump_rse_id, **did)
assert replica['tombstone'] == datetime(year=1970, month=1, day=1)
assert replica['state'] == ReplicaState.COPYING
# The intermediate replica is protected by its state (Copying)
rucio.daemons.reaper.reaper.REGION.invalidate()
reaper(once=True,
rses=[],
include_rses=jump_rse_name,
exclude_rses=None)
replica = replica_core.get_replica(rse_id=jump_rse_id, **did)
assert replica['state'] == ReplicaState.COPYING
# Wait for the intermediate replica to become ready
replica = __wait_for_replica_transfer(dst_rse_id=jump_rse_id, **did)
assert replica['state'] == ReplicaState.AVAILABLE
# The intermediate replica is protected by an entry in the sources table
# Reaper must not remove this replica, even if it has an obsolete tombstone
rucio.daemons.reaper.reaper.REGION.invalidate()
reaper(once=True,
rses=[],
include_rses=jump_rse_name,
exclude_rses=None)
replica = replica_core.get_replica(rse_id=jump_rse_id, **did)
assert replica
# FTS fails the second transfer, so run submitter again to copy from jump rse to destination rse
submitter(once=True,
rses=[{
'id': rse_id
} for rse_id in all_rses],
partition_wait_time=None,
transfertype='single',
filter_transfertool=None)
# Wait for the destination replica to become ready
replica = __wait_for_replica_transfer(dst_rse_id=dst_rse_id, **did)
assert replica['state'] == ReplicaState.AVAILABLE
rucio.daemons.reaper.reaper.REGION.invalidate()
reaper(once=True,
rses=[],
include_rses='test_container_xrd=True',
exclude_rses=None)
with pytest.raises(ReplicaNotFound):
replica_core.get_replica(rse_id=jump_rse_id, **did)
finally:
@transactional_session
def _cleanup_all_usage_and_limits(rse_id, session=None):
session.query(models.RSELimit).filter_by(rse_id=rse_id).delete()
session.query(models.RSEUsage).filter_by(
rse_id=rse_id, source='storage').delete()
_cleanup_all_usage_and_limits(rse_id=jump_rse_id)
def test_reaper():
""" REAPER (DAEMON): Test the reaper daemon."""
rse_name = rse_name_generator()
rse_id = rse_core.add_rse(rse_name)
mock_protocol = {
'scheme': 'MOCK',
'hostname': 'localhost',
'port': 123,
'prefix': '/test/reaper',
'impl': 'rucio.rse.protocols.mock.Default',
'domains': {
'lan': {
'read': 1,
'write': 1,
'delete': 1
},
'wan': {
'read': 1,
'write': 1,
'delete': 1
}
}
}
rse_core.add_protocol(rse_id=rse_id, parameter=mock_protocol)
nb_files = 30
file_size = 2147483648 # 2G
file_names = []
for i in range(nb_files):
file_name = 'lfn' + generate_uuid()
file_names.append(file_name)
replica_core.add_replica(rse_id=rse_id,
scope=InternalScope('data13_hip'),
name=file_name,
bytes=file_size,
tombstone=datetime.utcnow() -
timedelta(days=1),
account=InternalAccount('root'),
adler32=None,
md5=None)
rse_core.set_rse_usage(rse_id=rse_id,
source='storage',
used=nb_files * file_size,
free=800)
rse_core.set_rse_limits(rse_id=rse_id,
name='MinFreeSpace',
value=10737418240)
rse_core.set_rse_limits(rse_id=rse_id,
name='MaxBeingDeletedFiles',
value=10)
rses = [
rse_core.get_rse(rse_id),
]
reaper(once=True, rses=rses)
reaper(once=True, rses=rses)
assert_equal(
len(
list(
replica_core.list_replicas(dids=[{
'scope':
InternalScope('data13_hip'),
'name':
n
} for n in file_names],
rse_expression=rse_name))),
nb_files - 10)
def test_archive_removal_impact_on_constituents(rse_factory, did_factory, mock_scope, root_account, caches_mock, file_config_mock):
[cache_region] = caches_mock
rse_name, rse_id = rse_factory.make_mock_rse()
scope = mock_scope
account = root_account
# Create 2 archives and 4 files:
# - One only exists in the first archive
# - One in both, plus another replica, which is not in an archive
# - One in both, plus another replica, which is not in an archive; and this replica has expired
# - One in both, plus another replica, which is not in an archive; and this replica has expired; but a replication rule exists on this second replica
# Also add these files to datasets, one of which will be removed at the end
nb_constituents = 4
nb_c_outside_archive = nb_constituents - 1
constituent_size = 2000
archive_size = 1000
uuid = str(generate_uuid())
constituents = [{'scope': scope, 'name': 'lfn.%s.%d' % (uuid, i)} for i in range(nb_constituents)]
did_factory.register_dids(constituents)
c_first_archive_only, c_with_replica, c_with_expired_replica, c_with_replica_and_rule = constituents
replica_core.add_replica(rse_id=rse_id, account=account, bytes_=constituent_size, **c_with_replica)
replica_core.add_replica(rse_id=rse_id, account=account, bytes_=constituent_size,
tombstone=datetime.utcnow() - timedelta(days=1), **c_with_expired_replica)
replica_core.add_replica(rse_id=rse_id, account=account, bytes_=constituent_size,
tombstone=datetime.utcnow() - timedelta(days=1), **c_with_replica_and_rule)
rule_core.add_rule(dids=[c_with_replica_and_rule], account=account, copies=1, rse_expression=rse_name, grouping='NONE',
weight=None, lifetime=None, locked=False, subscription_id=None)
archive1, archive2 = [{'scope': scope, 'name': 'archive_%s.%d.zip' % (uuid, i)} for i in range(2)]
replica_core.add_replica(rse_id=rse_id, bytes_=archive_size, account=account, **archive1)
replica_core.add_replica(rse_id=rse_id, bytes_=archive_size, account=account, **archive2)
did_core.attach_dids(dids=[{'scope': c['scope'], 'name': c['name'], 'bytes': constituent_size} for c in constituents],
account=account, **archive1)
did_core.attach_dids(dids=[{'scope': c['scope'], 'name': c['name'], 'bytes': constituent_size} for c in [c_with_replica, c_with_expired_replica, c_with_replica_and_rule]],
account=account, **archive2)
dataset1, dataset2 = [{'scope': scope, 'name': 'dataset_%s.%i' % (uuid, i)} for i in range(2)]
did_core.add_did(did_type='DATASET', account=account, **dataset1)
did_core.attach_dids(dids=constituents, account=account, **dataset1)
did_core.add_did(did_type='DATASET', account=account, **dataset2)
did_core.attach_dids(dids=[c_first_archive_only, c_with_expired_replica], account=account, **dataset2)
@read_session
def __get_archive_contents_history_count(archive, session=None):
return session.query(ConstituentAssociationHistory).filter_by(**archive).count()
# Run reaper the first time.
# the expired non-archive replica of c_with_expired_replica must be removed,
# but the did must not be removed, and it must still remain in the dataset because
# it still has the replica from inside the archive
assert replica_core.get_replica(rse_id=rse_id, **c_with_expired_replica)
cache_region.invalidate()
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=2 * archive_size + nb_c_outside_archive * constituent_size)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=2 * archive_size + nb_c_outside_archive * constituent_size, free=1)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
for did in constituents + [archive1, archive2]:
assert did_core.get_did(**did)
for did in [archive1, archive2, c_with_replica, c_with_replica_and_rule]:
assert replica_core.get_replica(rse_id=rse_id, **did)
with pytest.raises(ReplicaNotFound):
# The replica is only on the archive, not on the constituent
replica_core.get_replica(rse_id=rse_id, **c_first_archive_only)
with pytest.raises(ReplicaNotFound):
# The replica outside the archive was removed by reaper
nb_c_outside_archive -= 1
replica_core.get_replica(rse_id=rse_id, **c_with_expired_replica)
# Compared to get_replica, list_replicas resolves archives, must return replicas for all files
assert len(list(replica_core.list_replicas(dids=constituents))) == 4
assert len(list(did_core.list_content(**dataset1))) == 4
assert len(list(did_core.list_archive_content(**archive1))) == 4
assert len(list(did_core.list_archive_content(**archive2))) == 3
assert __get_archive_contents_history_count(archive1) == 0
assert __get_archive_contents_history_count(archive2) == 0
# Expire the first archive and run reaper again
# the archive will be removed; and c_first_archive_only must be removed from datasets
# and from the did table.
replica_core.set_tombstone(rse_id=rse_id, tombstone=datetime.utcnow() - timedelta(days=1), **archive1)
cache_region.invalidate()
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=2 * archive_size + nb_c_outside_archive * constituent_size)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=2 * archive_size + nb_c_outside_archive * constituent_size, free=1)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
with pytest.raises(DataIdentifierNotFound):
assert did_core.get_did(**archive1)
with pytest.raises(DataIdentifierNotFound):
assert did_core.get_did(**c_first_archive_only)
assert len(list(replica_core.list_replicas(dids=constituents))) == 3
assert len(list(did_core.list_content(**dataset1))) == 3
assert len(list(did_core.list_archive_content(**archive1))) == 0
assert len(list(did_core.list_archive_content(**archive2))) == 3
assert __get_archive_contents_history_count(archive1) == 4
assert __get_archive_contents_history_count(archive2) == 0
# Expire the second archive replica and run reaper another time
# c_with_expired_replica is removed because its external replica got removed at previous step
# and it exists only inside the archive now.
# If not open, Dataset2 will be removed because it will be empty.
did_core.set_status(open=False, **dataset2)
replica_core.set_tombstone(rse_id=rse_id, tombstone=datetime.utcnow() - timedelta(days=1), **archive2)
cache_region.invalidate()
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=archive_size + nb_c_outside_archive * constituent_size)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=archive_size + nb_c_outside_archive * constituent_size, free=1)
reaper(once=True, rses=[], include_rses=rse_name, exclude_rses=None)
# The archive must be removed
with pytest.raises(DataIdentifierNotFound):
assert did_core.get_did(**archive2)
# The DIDs which only existed in the archive are also removed
with pytest.raises(DataIdentifierNotFound):
assert did_core.get_did(**c_first_archive_only)
with pytest.raises(DataIdentifierNotFound):
assert did_core.get_did(**c_with_expired_replica)
# If the DID has a non-expired replica outside the archive without rules on it, the DID is not removed
assert did_core.get_did(**c_with_replica)
# If the DID has an expired replica outside the archive, but has rules on that replica, the DID is not removed
assert did_core.get_did(**c_with_replica_and_rule)
assert len(list(replica_core.list_replicas(dids=constituents))) == 2
assert len(list(did_core.list_content(**dataset1))) == 2
with pytest.raises(DataIdentifierNotFound):
did_core.get_did(**dataset2)
assert len(list(did_core.list_content(**dataset2))) == 0
assert len(list(did_core.list_archive_content(**archive2))) == 0
assert __get_archive_contents_history_count(archive1) == 4
assert __get_archive_contents_history_count(archive2) == 3
def format_rses(rses_d, rses):
new_rses = {}
for rse in rses_d:
rse_name = rse
cric_data = rses_d[rse]
attributes_map = {
rse_name: "true",
"source_for_used_space": "rucio",
"fts": cric_data["fts"],
"verify_checksum": rses[rse]['verify_checksum'],
"lfn2pfn_algorithm": cric_data["lfn2pfn_algorithm"]
}
# add custom CRIC parameters for RSE
custom_params = cric_data['params']
for k, v in custom_params.items():
attributes_map[k] = v
# set default rse_type to DISK if not provided
if not cric_data["rse_type"]:
cric_data["rse_type"] = "DISK"
try:
rse_id = rse_module.get_rse_id(rse=rse_name)
total_space = cric_data["space"]
if total_space > 0:
rse_module.set_rse_usage(rse_id,
"storage",
-1,
total_space,
files=None)
rse_module.set_rse_usage(rse_id,
"obsolete",
0,
None,
files=None)
except RSENotFound:
# this means that the rse usage metrics will be updated in the next
# run for this RSE
pass
data = {
"MaxBeingDeletedFiles":
cric_data["MaxBeingDeletedFiles"],
"MinFreeSpace":
cric_data["MinFreeSpace"],
"availability_delete":
cric_data["availability_delete"],
"availability_read":
cric_data["availability_read"],
"availability_write":
cric_data["availability_write"],
"country_name":
cric_data["country_name"],
"deterministic":
cric_data["deterministic"],
"fts":
cric_data["fts"],
"impl":
cric_data["impl"],
"latitude":
cric_data["latitude"],
"lfn2pfn_algorithm":
cric_data["lfn2pfn_algorithm"],
"longitude":
cric_data["longitude"],
"region_code":
cric_data["region_code"],
"rse":
rse_name,
"rse_type":
cric_data["rse_type"],
"staging_area":
cric_data["staging_area"],
"timezone":
cric_data["timezone"],
"updated_at":
cric_data["updated_at"],
"volatile":
cric_data["volatile"],
# Missing from CRIC
"city":
"Missing from CRIC",
"availability":
"7",
"credentials":
"null",
"created_at":
"",
"verify_checksum":
rses[rse]['verify_checksum'],
"attributes":
attributes_map,
# Protocols
"protocols":
format_protocols(cric_data["protocols"], cric_data["impl"])
}
new_rses[rse_name] = data
return new_rses
def test_bb8_full_workflow(vo, root_account, jdoe_account, rse_factory,
mock_scope, did_factory):
"""BB8: Test the rebalance rule method"""
config_core.set(section='bb8', option='allowed_accounts', value='jdoe')
tot_rses = 4
rses = [rse_factory.make_posix_rse() for _ in range(tot_rses)]
rse1, rse1_id = rses[0]
rse2, rse2_id = rses[1]
rse3, rse3_id = rses[2]
rse4, rse4_id = rses[3]
# Add Tags
# RSE 1 and 2 nmatch expression T1=true
# RSE 3 and 4 nmatch expression T2=true
T1 = tag_generator()
T2 = tag_generator()
add_rse_attribute(rse1_id, T1, True)
add_rse_attribute(rse2_id, T1, True)
add_rse_attribute(rse3_id, T2, True)
add_rse_attribute(rse4_id, T2, True)
# Add fake weights
add_rse_attribute(rse1_id, "fakeweight", 10)
add_rse_attribute(rse2_id, "fakeweight", 0)
add_rse_attribute(rse3_id, "fakeweight", 0)
add_rse_attribute(rse4_id, "fakeweight", 0)
add_rse_attribute(rse1_id, "freespace", 1)
add_rse_attribute(rse2_id, "freespace", 1)
add_rse_attribute(rse3_id, "freespace", 1)
add_rse_attribute(rse4_id, "freespace", 1)
# Add quota
set_local_account_limit(jdoe_account, rse1_id, -1)
set_local_account_limit(jdoe_account, rse2_id, -1)
set_local_account_limit(jdoe_account, rse3_id, -1)
set_local_account_limit(jdoe_account, rse4_id, -1)
set_local_account_limit(root_account, rse1_id, -1)
set_local_account_limit(root_account, rse2_id, -1)
set_local_account_limit(root_account, rse3_id, -1)
set_local_account_limit(root_account, rse4_id, -1)
# Invalid the cache because the result of parse_expression is cached
REGION.invalidate()
tot_datasets = 4
# Create a list of datasets
datasets = [did_factory.make_dataset() for _ in range(tot_datasets)]
dsn = [dataset['name'] for dataset in datasets]
rules = list()
base_unit = 100000000000
nb_files1 = 7
nb_files2 = 5
nb_files3 = 3
nb_files4 = 2
file_size = 1 * base_unit
rule_to_rebalance = None
# Add one secondary file
files = create_files(1, mock_scope, rse1_id, bytes_=1)
add_rule(dids=[{
'scope': mock_scope,
'name': files[0]['name']
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
for cnt in range(3, tot_rses):
add_replicas(rses[cnt][1], files, jdoe_account)
add_rule(dids=[{
'scope': mock_scope,
'name': files[0]['name']
}],
account=jdoe_account,
copies=1,
rse_expression=rses[cnt][0],
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
# Create dataset 1 of 800 GB and create a rule on RSE 1 and RSE 3
files = create_files(nb_files1, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[0], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[0]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse3_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[0]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
# Create dataset 2 of 500 GB and create a rule on RSE 1 and RSE 2
files = create_files(nb_files2, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[1], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[1]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse2_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[1]
}],
account=jdoe_account,
copies=1,
rse_expression=rse2,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
# Create dataset 3 of 300 GB and create a rule on RSE 1. The copy on RSE 3 is secondary
files = create_files(nb_files3, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[2], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[2]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rule_to_rebalance = rule_id
rules.append(rule_id)
add_replicas(rse3_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[2]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
try:
rule = get_rule(rule_id)
except:
pytest.raises(RuleNotFound, get_rule, rule_id)
# Create dataset 4 of 200 GB and create a rule on RSE 3. The copy on RSE 2 is secondary
files = create_files(nb_files4, mock_scope, rse3_id, bytes_=file_size)
attach_dids(mock_scope, dsn[3], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[3]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse2_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[3]
}],
account=jdoe_account,
copies=1,
rse_expression=rse2,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
try:
rule = get_rule(rule_id)
except:
pytest.raises(RuleNotFound, get_rule, rule_id)
for dataset in dsn:
set_status(mock_scope, dataset, open=False)
for rse in rses:
fill_rse_expired(rse[1])
set_rse_usage(rse_id=rse[1],
source='min_free_space',
used=2 * base_unit,
free=2 * base_unit,
session=None)
set_rse_usage(rse_id=rse[1],
source='storage',
used=15 * base_unit,
free=2 * base_unit,
session=None)
set_rse_usage(rse_id=rse2_id,
source='min_free_space',
used=1 * base_unit,
free=1 * base_unit,
session=None)
set_rse_usage(rse_id=rse2_id,
source='storage',
used=6 * base_unit,
free=5 * base_unit,
session=None)
run_abacus(once=True, threads=1, fill_history_table=False, sleep_time=10)
# Summary :
# RSE 1 : 1500 GB primary + 1 B secondary
tot_space = [
src for src in get_rse_usage(rse1_id) if src['source'] == 'rucio'
][0]
expired = [
src for src in get_rse_usage(rse1_id) if src['source'] == 'expired'
][0]
assert tot_space['used'] == (nb_files1 + nb_files2 +
nb_files3) * file_size + 1
assert expired['used'] == 1
# RSE 2 : 500 GB primary + 100 GB secondary
tot_space = [
src for src in get_rse_usage(rse2_id) if src['source'] == 'rucio'
][0]
expired = [
src for src in get_rse_usage(rse2_id) if src['source'] == 'expired'
][0]
assert tot_space['used'] == (nb_files2 + nb_files4) * file_size
assert expired['used'] == nb_files4 * file_size
# Total primary on T1=true : 2000 GB
# Total secondary on T1=true : 200 GB
# Ratio secondary / primary = 10 %
# Ratio on RSE 1 : 0 %
# Ratio on RSE 2 : 40 %
# Now run BB8
re_evaluator(once=True, sleep_time=30, did_limit=100)
bb8_run(once=True,
rse_expression='%s=true' % str(T1),
move_subscriptions=False,
use_dump=False,
sleep_time=300,
threads=1,
dry_run=False)
for rule_id in rules:
rule = get_rule(rule_id)
if rule_id != rule_to_rebalance:
assert (rule['child_rule_id'] is None)
else:
assert (rule['child_rule_id'] is not None)
assert (
rule['expires_at'] <= datetime.utcnow() + timedelta(seconds=1)
) # timedelta needed to prevent failure due to rounding effects
child_rule_id = rule['child_rule_id']
child_rule = get_rule(child_rule_id)
assert (child_rule['rse_expression'] == rse2)
# For teardown, delete child rule
update_rule(child_rule_id, {'lifetime': -86400})
rule_cleaner(once=True)
for dataset in dsn:
set_metadata(mock_scope, dataset, 'lifetime', -86400)
undertaker.run(once=True)
def test_reaper():
""" REAPER2 (DAEMON): Test the reaper2 daemon."""
if config_get_bool('common', 'multi_vo', raise_exception=False, default=False):
vo = {'vo': config_get('client', 'vo', raise_exception=False, default='tst')}
new_vo = {'vo': 'new'}
if not vo_core.vo_exists(**new_vo):
vo_core.add_vo(description='Test', email='*****@*****.**', **new_vo)
if not scope_core.check_scope(InternalScope('data13_hip', **new_vo)):
scope_core.add_scope(InternalScope('data13_hip', **new_vo), InternalAccount('root', **new_vo))
nb_rses = 2
else:
vo = {}
new_vo = {}
nb_rses = 1
mock_protocol = {'scheme': 'MOCK',
'hostname': 'localhost',
'port': 123,
'prefix': '/test/reaper',
'impl': 'rucio.rse.protocols.mock.Default',
'domains': {
'lan': {'read': 1,
'write': 1,
'delete': 1},
'wan': {'read': 1,
'write': 1,
'delete': 1}}}
nb_files = 250
file_size = 200 # 2G
rse_names = []
all_file_names = []
for j in range(nb_rses):
rse_name = rse_name_generator()
rse_names.append(rse_name)
rse_id = rse_core.add_rse(rse_name, **vo)
rse_core.add_protocol(rse_id=rse_id, parameter=mock_protocol)
if new_vo:
rse_id_new = rse_core.add_rse(rse_name, **new_vo)
rse_core.add_protocol(rse_id=rse_id_new, parameter=mock_protocol)
file_names = []
for i in range(nb_files):
file_name = 'lfn' + generate_uuid()
file_names.append(file_name)
replica_core.add_replica(rse_id=rse_id, scope=InternalScope('data13_hip', **vo),
name=file_name, bytes=file_size,
tombstone=datetime.utcnow() - timedelta(days=1),
account=InternalAccount('root', **vo), adler32=None, md5=None)
if new_vo:
replica_core.add_replica(rse_id=rse_id_new, scope=InternalScope('data13_hip', **new_vo),
name=file_name, bytes=file_size,
tombstone=datetime.utcnow() - timedelta(days=1),
account=InternalAccount('root', **new_vo), adler32=None, md5=None)
all_file_names.append(file_names)
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=1)
rse_core.set_rse_limits(rse_id=rse_id, name='MinFreeSpace', value=50 * file_size)
# rse_core.set_rse_limits(rse_id=rse_id, name='MaxBeingDeletedFiles', value=10)
if new_vo:
rse_core.set_rse_usage(rse_id=rse_id_new, source='storage', used=nb_files * file_size, free=1)
rse_core.set_rse_limits(rse_id=rse_id_new, name='MinFreeSpace', value=50 * file_size)
# rse_core.set_rse_limits(rse_id=rse_id_new, name='MaxBeingDeletedFiles', value=10)
from rucio.daemons.reaper.reaper2 import REGION
REGION.invalidate()
if not vo:
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **vo), 'name': n} for n in all_file_names[0]],
rse_expression=rse_name))) == nb_files
# Check first if the reaper does not delete anything if no space is needed
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=323000000000)
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **vo), 'name': n} for n in all_file_names[0]],
rse_expression=rse_name))) == nb_files
# Now put it over threshold and delete
rse_core.set_rse_usage(rse_id=rse_id, source='storage', used=nb_files * file_size, free=1)
from rucio.daemons.reaper.reaper2 import REGION
REGION.invalidate()
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **vo), 'name': n} for n in all_file_names[0]],
rse_expression=rse_name))) == 200
else:
# Check we reap all VOs by default
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
reaper(once=True, rses=[], include_rses=rse_names[0], exclude_rses=[])
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **vo), 'name': n} for n in all_file_names[0]],
rse_expression=rse_names[0]))) == 200
assert len(list(replica_core.list_replicas(dids=[{'scope': InternalScope('data13_hip', **new_vo), 'name': n} for n in all_file_names[0]],
rse_expression=rse_names[0]))) == 200
